Abstract:Western metadata often flattens the dynamic "Qi" of Traditional Chinese Patterns. Utilizing a 27,000-item annotated dataset, we construct a "Subject-Motion-Context" ontology to encode indigenous visual syntax. This project challenges static classification systems, restoring narrative sovereignty and providing a dynamic semantic framework for AI-driven cultural engagement.

Keywords:Traditional Chinese Patterns; Indigenous Data Sovereignty; Dynamic Ontology; Knowledge Graph; Semantic Semiotics

1. Introduction: The Silent "Memory Machines" and the Gap in Engagement

The theme of DH2026, "Engagement," calls for a critical re-examination of how digital technologies mediate the relationship between human communities and their cultural heritage. In the context of the "Memory of the World," archives are not merely static repositories of artifacts but dynamic systems of meaning-making. However, as the digital humanities increasingly embrace global perspectives, a fundamental epistemological gap remains: established digital standards for cultural heritage—such as Iconclass, the Art & Architecture Thesaurus (AAT), or even the generic labeling of Computer Vision (CV) datasets—are predominantly rooted in Western art historical classifications. These systems tend to prioritize static taxonomies (identifying what an object is) over dynamic relations (describing how an object acts).

For Traditional Chinese Patterns (TCP), this static approach represents a form of "digital forgetting." In the Chinese aesthetic tradition, a pattern is never just a noun; it is a visual manifestation of "Qi" (Spirit Resonance/Vitality). A "Dragon" is not a singular, monolithic category; its cultural meaning shifts entirely depending on whether it is "Sitting" (symbolizing consolidated imperial power), "Soaring" (symbolizing rising auspicious energy), or "Walking" (symbolizing progression). Current "Memory Machines"—including Large Language Models and generative AI—fail to capture these nuanced syntactical rules, often reducing rich indigenous narratives to flattened, decorative stereotypes.

This paper critiques this representational deficit and proposes a transformative solution: a Chinese Pattern Description Model (CPDM). Grounded in a massive, manually annotated dataset, we construct a domain-specific ontology that encodes the "indigenous grammar" of pattern naming. By shifting the unit of analysis from the "image tag" to the "semantic sentence," we aim to restore the narrative sovereignty of these cultural archives.

2. Data Foundation: A Panorama of Myth and Semantics

The empirical basis of this study is a newly constructed, comprehensive dataset of Traditional Chinese Patterns, curated from authoritative archaeological reports, historical design manuals, and museum catalogs.

2.1 Scale and Composition

The dataset comprises over 27,000 individual semantic units, manually annotated to capture deep semantic relationships. The data structure is not limited to visual labels but includes rich metadata fields: Subject Name, Category, Thematic Meaning, Constituent Elements, Carrier (Artifact), Ethnicity, Region, Dynasty, and Historical Context.

2.2 The "Mythical" Nature of the Archive

A distinctive feature of this dataset is its distribution, which mirrors the collective imagination of Chinese civilization rather than the physical world.

Mythical/Divine Patterns constitute the largest category (23.5%, n=5,202), significantly outnumbering Natural Patterns (3.9%).

Within this category, the dataset documents 1,479 variations of the Dragon (Long) and 1,380 variations of the Phoenix (Feng).

Other major categories include Plants (23.1%), Animals (21.4%), Geometry (11.7%), and Human Figures (8.2%)

This distribution presents a unique challenge for digital analysis: these are not representations of natural objects that can be easily identified by standard object-detection algorithms. They are "Constructed Symbols" governed by strict cultural rules. The dataset documents 402 distinct pattern names, ranging from the common "Cloud Pattern" to the complex "Dragon Passing Through Peonies." This granularity provides the necessary corpus to decode the linguistic rules of visual design.

3. Methodology: From Naming Conventions to Computable Ontology

Our core methodology bridges linguistics and computer science. We treat the traditional naming conventions of Chinese patterns not as arbitrary labels, but as a rigorous "Visual Grammar."

3.1 Syntactic Analysis: The Four-Dimensional Formula

Through the linguistic deconstruction of the 402 pattern names and their corresponding 27,000 descriptions, we identified a consistent syntactic structure. We formalize this as a semantic function:

Pattern(P) = f(Quantity(Q) + Motion(M) + Subject(S) + Combination(C)) → Meaning(I)

This formula challenges the standard "Subject-Predicate" logic of Western metadata.

Subject (S): The core element (e.g., Dragon, Lotus).

Quantity (Q): A semantic multiplier (e.g., "Nine" represents the supreme, "Double" represents marital bliss).

Motion (M): The dynamic state (e.g., Soaring, Looking Back).

Combination (C): The topological relationship (e.g., Intertwined, Surrounded).

3.2 Ontology Construction: Encoding "Motion" (The Variable M)

The most critical innovation of our ontology is the formalization of "Motion Vectors." In standard thesauri, a dragon is simply a Dragon. In our CPDM ontology, we mapped specific verbs found in the dataset to hierarchical cultural meanings:

"Zuo" (Sitting/Squatting): Found in patterns like "Sitting Dragon" or "Sitting Lion." This state maps to semantics of Stability, Dignity, and Authority.

"Teng" (Soaring/Ascending): Found in "Soaring Dragon." Maps to High Ambition, Rising Power, and Divine Intervention.

"Xing" (Walking): Maps to Grace, Progression, and Rhythm.

"Xi" (Playing/Teasing): Found in "Dragon Playing with Pearl." Maps to Interaction, Celebration, and Vitality.

By encoding these verbs as properties (hasState or performsAction) within the ontology, we ensure that the digital surrogate retains the specific ritual function of the original artifact.

3.3 The Logic of Combination: Topology as Meaning

The ontology also defines rules for element combinations (C).

The "Chanzhi" (Intertwined) Rule: When a plant subject is modified by the "Intertwined" structure (e.g., Intertwined Lotus, n=786), the semantic meaning shifts from simply "Purity" (Lotus) to "Endless Continuity and Lineage."

The "Chuan" (Passing Through) Rule: When a Mythical Beast (Subject A) combines with a Plant (Subject B) via the "Passing Through" topology (e.g., Dragon Passing Through Peonies), the ontology infers a composite meaning: The power of the Dragon protects and engages with the wealth symbolized by the Peony.

We implemented this model using OWL (Web Ontology Language) and RDF (Resource Description Framework), creating a Knowledge Graph that allows for inferential querying (e.g., "Find all patterns that symbolize 'Lineage' through 'Plant Morphology'").

4. Case Studies and Critical Analysis

We applied the CPDM to analyze specific subsets of the data, revealing patterns of "Collective Memory" that were previously invisible in unstructured archives.

Case Study 1: The Evolution of the "Dragon" Narrative

Analyzing the 1,479 "Dragon" entries, we traced the temporal evolution of the Motion Variable.

In early dynasties (e.g., Han), "Walking" (Xing) and "Running" (Ben) states were predominant, reflecting a cultural emphasis on physical vigor and animism.

In later dynasties (e.g., Ming/Qing), "Sitting" (Zuo) and frontal-facing "Soaring" (Teng) patterns became dominant (accounting for over 60% of imperial dragon patterns in the dataset).

Interpretation: This shift quantifies the fossilization of political power. The ontology makes visible how the "Memory of the Dragon" transformed from a wild spirit to a ritualized symbol of statecraft.

Case Study 2: The "Homophonic" Memory Mechanism

The dataset contains 68 entries for "Bat" (Bianfu). In a CV model, these are classified visually as "Chiroptera." However, our ontology links the Subject Bat to the Phonetic Value Fu, which connects to the Semantic Concept Good Fortune. Furthermore, when the dataset detects the Quantity Five + Subject Bat (n=29), the ontology automatically infers the "Five Blessings" (Wu Fu). This demonstrates how the CPDM captures the "textual memory" embedded within the "visual image."

5. Discussion: Towards Indigenous Data Sovereignty

This research directly addresses the "Engagement" theme by engaging with the politics of digital representation.

5.1 Challenging the "Memory Machine"

Generative AI models (like Midjourney) are currently the most powerful "Memory Machines." However, they often suffer from "Cultural Hallucinations"—generating Chinese-style images that look superficially correct but are grammatically nonsensical (e.g., a "Sitting Dragon" engaged in a "Flying" background). Our structured dataset and ontology provide the necessary "Ground Truth" to align these models. By teaching machines the grammar of patterns (Subject+Motion+Combination) rather than just pixel correlations, we can guide AI to generate culturally coherent content.

5.2 Decolonizing the Archive

By adopting an ontology structure derived from the indigenous naming practices of Chinese artisans (the "Verb-Object" structure) rather than importing Western art history standards, this project practices Indigenous Data Sovereignty. It asserts that the way we describe a cultural object is as important as the object itself. We preserve not just the image of the pattern, but the logic used by the culture to categorize its own world.

6. Conclusion

The "Grammar of Qi" project demonstrates that Digital Humanities can do more than digitize; it can reconstruct the cognitive frameworks of the past. By leveraging a high-quality, deeply annotated dataset of 27,000 items, we have built a bridge between the fluidity of traditional Chinese visual language and the structured rigidity of computational ontologies.

This work offers a scalable methodology for World Memory projects, suggesting that the key to meaningful engagement lies in respecting the "native syntax" of cultural archives. In the era of AI, preserving this syntax is the only way to ensure that our digital memories remain meaningful, diverse, and truly connected to their human origins.

1.Korzun, Dmitry / Varfolomeyev, Aleksey / Yalovitsyna, Svetlana / Volokhova, Valentina (2017): “Semantic infrastructure of a smart museum: toward making cultural heritage knowledge usable and creatable by visitors and professionals”, in: Personal and Ubiquitous Computing 21, 2: 345–354.

2.Guarino, Nicola / Masolo, Claudio / Vetere, Guido (1999): “OntoSeek: Content-Based Access to the Web”, in: IEEE Intelligent Systems 14, 3: 70–80.

3.Lewis, Patrick / Perez, Ethan / Piktus, Aleksandra / Petroni, Fabio / Karpukhin, Vladimir / 4.Goyal, Naman / Küttler, Heinrich / Lewis, Mike / Yih, Wen-tau / Rocktäschel, Tim / Riedel, 5.Sebastian / Kiela, Douwe (2020): “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks”, in: arXiv preprint arXiv:2005.11401.

6.Cobley, Paul (2010): The Routledge companion to semiotics. New York: Routledge.

7.Saussure, Ferdinand de (1966): Course in general linguistics. New York: McGraw-Hill.

8.Pérez, Gómez / Benjamins, V. Richard (1999): “Overview of knowledge sharing and reuse components: ontologies and problem-solving methods”, in: Proceedings of the IJCAI'99 Workshop on Ontologies and Problem-Solving Methods, Stockholm, August 1999. Stockholm: CEUR.

9.Schneider, Edward W. (1973): “Course Modularization Applied: The Interface System and Its Implications For Sequence Control and Data Analysis”, in: Proceedings of the 1973 Conference on Computers in the Undergraduate Curricula 2: 10–15.

10.Gu, Jindong / Han, Zhen / Chen, Shuo / Beirami, Ahmad / He, Zhen / Zhang, Chao / Liao, Rui / Young, Peter / Müller, Henrick / Gu, Shuai (2023): “A systematic survey of prompt engineering on vision-language foundation models”, in: arXiv preprint arXiv:2307.12980.

11.Rawson, Jessica (2006): “Ornament as system: Chinese bird-and-flower design”, in: The Burlington Magazine 148, 1239: 380–389.

12.Kim, Do Yeong (2020): “The Changing Nature and Significance of the Chinese-style Belt Ornament in East Asia”, in: Hanguk Kogo-Hakbo 115: 124–165.